Fix a couple of output/upload value issues in pycbc live #4953
Conversation
GarethCabournDavies
added
low latency
v27_release_branch
|
@GarethCabournDavies would it be possible for you to get single detector candidate info into the foreground section of the 8 second trigger files as well? or too much for this one MR? |
Probably too much for this MR (and avoiding scope-creep of this one). I'll add an issue to say that a lot of this (saving/upload of candidates) seems to be repeated imperfectly between coincs and singles, so trying to get them to use the same code path would be better |
Ok. Getting the likelihood into the xml is already an improvement, thanks! |
| @@ -303,7 +303,8 @@ def check(self, trigs, data_reader): | |||
|
|
|||
| # fill in a new candidate event | |||
| candidate = { | |||
| f'foreground/{self.ifo}/{k}': cutall_trigs[k][i] for k in trigs | |||
| f'foreground/{self.ifo}/{k}': cut_trigs[k][sngl_idx][i] | |||
There was a problem hiding this comment.
I'm a bit confused by the change to this line? Why do we need sngl_idx here now when we didn't before?
There was a problem hiding this comment.
cutall_trigs already had the sngl_idx included (line 285 above), but cut_trigs doesnt have that
There was a problem hiding this comment.
Ah I missed the change from cutall_trigs to cut_trigs
| @@ -303,7 +303,8 @@ def check(self, trigs, data_reader): | |||
|
|
|||
| # fill in a new candidate event | |||
| candidate = { | |||
| f'foreground/{self.ifo}/{k}': cutall_trigs[k][i] for k in trigs | |||
| f'foreground/{self.ifo}/{k}': cut_trigs[k][sngl_idx][i] | |||
There was a problem hiding this comment.
Ah I missed the change from cutall_trigs to cut_trigs
Update the chisquared uploaded to gracedb to be reduced, and save the ranking statistic value
Standard information about the request
This is a bug fix
This change affects the live search
This change changes scientific output
This change: follows style guidelines (See e.g. PEP8), has been proposed using the contribution guidelines
Motivation / contents
We noticed that the chisquared for singles was not the reduced chisquared - I tracked it down to where I'd missed that I was using the same trigger set for statistic calculation as I was for the upload, which is not what is done for coincs, this is now set back to the un-modified set of triggers.
We also realised that the ranking statistic isnt being saved - I have added this to the gracedb upload/xml by utilising the "likelihood" value in the coinc row of the table.
Testing performed
Need to test this, but thought it best to get the PR in first